Introduction

The management of Crohn’s disease (CD) has significantly changed over the past two decades. Whereas, in the past, surgery was considered to be the mainstay of treatment, immunosuppressive medications that target inflammatory pathways have emerged as the cornerstone of contemporary management of CD.1 Despite this shift towards more conservative treatment modalities, most patients with CD will eventually require an operative procedure at some point in their future.2,3,4,5

The key pharmacological agents employed in the current management of CD namely corticosteroids, biologics, and immunomodulators have been reported to be associated with increased rates of postoperative infectious complications probably as a consequence of their immunomodulating effect which targets the same pathways involved in wound healing and pathogen clearance.6,7,8,9 Conversely, the association of infectious complications and immunosuppressive agents has been refuted by other studies.10,11 Notwithstanding the large body of existing literature that has attempted to address this question, there remains much equipoise regarding the impact of biologic agents on postoperative outcomes in inflammatory bowel diseases (IBD). This is especially apparent since the majority of reported studies that have attempted to address this issue have been mostly non-randomized, underpowered small-retrospective studies, which limited conclusions that could be drawn from systematic reviews and meta-analyses.8,12,13,14,15,16 Thus, the aim of the present study was to evaluate the impact of preoperative immunosuppression on the development of postoperative infectious complications following elective colorectal resections in CD patients using a large prospective validated database.

Materials and Methods

Study Population

The study population consisted of adults (≥ 18 years of age) with CD who underwent elective colon resections from 2011 to 2015 from the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP). ACS-NSQIP is a nationally validated prospective, comprehensive database made up of over 400 institutions aimed to improve the quality of surgical care. This study received institutional ethics review board approval.

Inclusion Criteria

Patients with CD were identified according to International Classification of Diseases, Ninth Revision (ICD-9) codes 555.0, 555.1, 555.2, and 555.9. CD patients who underwent colon resections were captured using the appropriate Current Procedural Terminology (CPT) codes (Table 1). The ACS-NSQIP definition for immunosuppression in IBD further subdivided these patients into immunosuppressed (IMS) or immunocompetent (IMC). According to this definition, individuals who received intravenous or oral corticosteroids for more than 10 days or immunosuppressant medications in the 30 days prior to surgery or at evaluation for surgery for IBD were considered immunosuppressed. Immunosuppressant medications included in this definition were mycophenolate mofetil, adalimumab, etanercept, azathioprine, cyclosporine, tacrolimus, sirolimus, infliximab, natalizumab, methotrexate, and certolizumab pegol.

Table 1 CPT codes for patients who underwent colon resections in this study
Full size table

Exclusion Criteria

Patients who underwent an emergency procedure were excluded. Emergency surgery was defined by ACS-NSQIP as operations that occurred within a short interval of time between diagnosis or symptomatology implying the patient’s well-being was threatened by the unnecessary delay of surgical intervention. Also excluded were patients who received chemotherapy within 90 days of surgery, patients who had evidence of preoperative septic shock or intubated preoperatively, patients with an American Society of Anesthesiologists (ASA) classification 5, and patients with disseminated cancer.

Outcomes

The primary outcome was infectious complications, a composite outcome defined as all surgical site infection (SSI), urinary tract infection, pneumonia, sepsis, and septic shock. Overall SSI (superficial, deep, and organ space surgical site infection), organ space infection surgical site infection, and anastomotic leak comprised secondary outcomes. Superficial, deep, and organ space infections surgical site infection (OSI) were defined by NSQIP as an occurrence of infection that manifests in the skin or subcutaneous tissues of the incision, deep soft tissues of the incision and organs or spaces other than the incision, which was opened or manipulated during the operation, respectively. Reoperation was defined by NSQIP as unplanned return to the operating room within 30 days of surgery for any surgical procedure.

Statistical Analysis

On univariate analysis, normally distributed and categorical variables were compared using a two-tailed Student’s t test and Pearson’s chi-square test, respectively. We used Wilcoxon rank sum test for continuous variables that were not normally distributed. Patients with significant missing data (> 20%) were excluded from the analysis. The association between immunosuppression and risk of postoperative complications was evaluated by means of multivariate logistic regression. The regression models were adjusted for known confounders as well as variables found to be statistically significant on univariate analysis. A two-tailed p value less than 0.05 was considered to indicate statistical significance. Statistical analyses were performed using Stata software version14.1 (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP.)

Results

Of the 94,055 patients in the colectomy-specific ACS-NSQIP database from 2011 to 2015, a total of 3860 patients with CD required elective colon resections and met the inclusion criteria (Fig. 1). Of those, 2483 (64.33%) patients were IMS and 1377 patients (35.67%) were IMC. Preoperative and demographic variables in both groups are detailed in Table 2.

Fig. 1
figure 1

Patients who met inclusion criteria (IMS, immunosuppressed; IMC, immunocompetent; PUF, patient user file)

Full size image
Table 2 Univariate comparison of preoperative and operative characteristics of IMS vs. IMC patients with CD
Full size table

On univariate analysis, there were significant differences amongst both groups. IMS patients had increased rates of infectious complications (16.23% vs. 13.58%, p = 0.028), SSI (13.09% vs. 9.94% p = 0.004), OSI (7.17% vs. 4.79%, p = 0.004), and anastomotic leak (4.07% vs. 2.55%, p = 0.014) (Table 3). The odds of infectious complications on multivariate logistic regression were 25% higher for IMS compared to IMC patients (odds ratio, OR 1.25; 95% CI 1.033–1.523) (Table 4). The increase in odds in IMS relative to IMC patients remained true for all SSI (OR 1.49; 1.128–1.742), OSI (OR 1.47; 1.094–1.984), and anastomotic leak (OR 1.51; 1.018–2.250) (Tables 5, 6, and 7). In addition to immunosuppression, the multivariate regressions performed identified other independent predictors of postoperative complications including open surgery, smoking status, and wound contamination (Tables 5, 6, and 7). In a subgroup of 3216 patients with an anastomosis and without any diversion, the odds of infectious complications, all SSI, and OSI (1.28, 1.31, and 1.38, respectively) remained significantly greater for IMS compared to IMC; however, the anastomotic leak rate was not significantly greater in IMS (OR 1.26, 95% CI 0.86 to 1.98).

Table 3 Univariate comparison of outcomes in IMS vs. IMC patients with CD following elective colon resections
Full size table
Table 4 Multivariate logistic regression for infectious complications following elective surgery in IMS patients with CD
Full size table
Table 5 Multivariate logistic regression for all SSI following elective surgery in IMS patients with CD
Full size table
Table 6 Multivariate logistic regression for organ space SSI following elective surgery in IMS patients with CD
Full size table
Table 7 Multivariate logistic regression for anastomotic leak following elective surgery in IMS patients with CD
Full size table

Discussion

In the present study, a significant association between immunosuppression for CD and postoperative infectious complications was identified in a cohort of 3860 CD patients who underwent elective colectomies. This increased risk was observed for a composite outcome of infectious complications, all SSI, OSI, and anastomotic leak. In fact, the odds of all SSI were 40% greater in IMS compared to IMC patients, which could be explained by the finding that the odds of OSI were 47% higher in IMS CD patients. Likewise, a parallel increased risk of anastomotic leak was also observed in IMS compared to IMC patients. The results of increased risk of OSI and anastomotic leak are clinically significant as these infections result in morbidity, prolong hospital length of stay, and often necessitate re-intervention. These findings were observed despite controlling for important confounders including age, body mass index, preoperative weight loss, diabetes, smoking, chronic obstructive pulmonary disease, hypertension, wound contamination, laparoscopic approach, presence of a stoma, and type of colonic resection.

The assumption that immunosuppressant medications may predispose to postoperative complications is rooted in their role in inhibiting the inflammatory cascade, a key component of the wound healing process and remodeling phase.17 Although several studies have examined the association between immunosuppression and postoperative complications in patients with CD, the literature is fraught with heterogeneity with regard to exposure to different immunosuppressive agents. Using results from a longitudinal CD database of ileocolic resections, Appau et al.18 compared 60 patients who had received infliximab to 389 controls. The authors observed a heightened risk of intra-abdominal sepsis, intra-abdominal abscess, anastomotic leak, and readmissions in those patients who had received anti-TNF-α prior to surgery compared to those who had not. Other studies, however, failed to demonstrate a significant association between perioperative treatment with biological agents and/or other immunosuppressants and postoperative complications.17,19,20,21 In a retrospective review of 413 patients with CD, UC, or indeterminate colitis who underwent a variety of abdominal procedures, preoperative infliximab (n = 101) was not found to increase crude postoperative complications (16.8% vs. 15.7%, p = 1). Similarly, on multivariate regression, the use of infliximab (OR 2.5, p = 0.14) or steroids (OR 1.2, p = 0.74) was not shown to be associated with increased risk of postoperative infections after adjusting for potential confounders.19 In addition, a case-control study that matched patients on operative procedure, IBD subtype, exposure to steroids, and patient age (195 cases on anti-TNF-α and 278 matched controls) did not find any difference between the two groups with regard to postoperative outcomes on multivariate analysis.21 In another study, Canedo et al.10 assessed 225 patients with CD reviewing their comorbidities, type of operations, and postoperative complications. Patients were grouped based on having received immunosuppressive medications within 90 days of the surgery (infliximab group, group having received steroids and other immunosuppressive drugs, and controls receiving no drugs). The authors did not observe a statistically significant difference in the rates of postoperative infectious complications amongst the groups. Furthermore, a recent meta-analysis that included 21 studies comparing CD patients who received preoperative immunosuppressive agents with those who had not clearly highlighted the heterogeneity of available data.8 Of these, six studies compared outcomes for patients with CD who received a varied combination of immunosuppressive agents including steroids, immunomodulators, and anti-TNF-α medications. In this subgroup of 1264 patients, no significant difference in infectious complications or overall morbidity was observed.8 However, when the authors limited the analysis to studies that included exposure to only anti-TNF-α (n = 14 studies, RR 1.29; 95% CI 1.07–1.55) or steroids (n = 13 studies, OR 1.55; 95% CI 1.23–1.95), they observed an increased risk of infectious complications. As noted by the authors of the review, the quality of included studies remained suboptimal as most were underpowered, retrospective, and unmatched. Although a randomized controlled trial would be ideal in resolving this issue, such a study design would be difficult due to the required large sample size and obvious barriers in the feasibility of randomizing critically symptomatic patients with CD who are awaiting surgery. Thus, the use of a large prospectively validated dataset such as the ACS-NSQIP is helpful in filling some of these gaps.

In a recent publication by Valizadeh et al.22 looking at the impact of immunosuppression on 30-day postoperative outcomes for patients with CD who underwent emergency or elective colectomy using the ACS-NSQIP database for the years 2012–2013, the authors found that IMS patients were at increased risk of anastomotic leak, sepsis, and septic shock. Despite the significance of these results, the data are difficult to translate into clinical practice as this heterogeneous cohort included emergent and elective colectomies, without taking diversion or type of resection into consideration. Many patients who undergo emergency surgery for CD are significantly more ill than elective patients, have preoperative intra-abdominal sepsis, have more extensive resections, and are most often best managed with concomitant diversion.1,23,24,25 In fact, emergency surgery for IBD has been repeatedly shown to be a significant risk factor for such complications and death.26,27 For this reason, in the present study, we chose to include a larger cohort of patients (from 2012 to 2015) and to limit the study group to a more homogeneous CD population, without decreasing the power of the hypothesis testing. Indeed, to create a clear and homogeneous cohort from which conclusions applicable to clinical practice could be obtained, patients with evidence of emergency surgery, preoperative septic shock, intubation, or ASA 5 scores, in addition to patients with preoperative chemotherapy/radiotherapy or disseminated cancer, were excluded.

Interestingly, in addition to the significantly greater risk of infectious complications observed in the IMS group, we observed that the types of operations performed differed in both groups. IMS patients were observed to have had significantly greater rates of subtotal colectomies and concomitant stomas compared to IMC patients who had increased rates of segmental colectomies. This is intuitive as the patients on immunosuppressive medications may have had more advanced disease requiring a total colectomy, diversion, or no anastomosis, or it may reflect surgeons’ reluctance to create an anastomosis in this immunosuppressed population. In addition, we observed higher rates of laparoscopy for IMS compared to IMC patients as did Canedo et al.10 They described a higher rate of laparoscopic surgery in the group that was treated with infliximab, steroids, or other immunosuppressive agents compared to the group that received no drugs (47.7% vs. 45.9% vs. 29.3% p = 0.04). This may reflect referral bias as IMS patients may be more likely to be referred to tertiary care centers with advanced minimally invasive colorectal expertise. Given that IMS patients are at increased odds of infectious complications, laparoscopy, an independent protective factor for the latter, may help mitigate significant morbidity in this subgroup.28,29 Thus, the importance of a minimally invasive approach in minimizing complications following resections for patients with CD should be highlighted. Moreover, risk factors for complications identified such as smoking should be considered when counseling patients for surgery and for preoperative risk adjustment.

The major strength of this study lies in its use of a large, multi-institutional nationally validated database that allows for increased generalizability. The quality of this dataset has been repeatedly validated through regular stringent audits, making the longitudinal database a strong base for hypothesis testing. Moreover, due to the breadth of available data within the database and the large sample size, many possible confounders could be accounted for in the multivariate analysis. In addition, the individual effect of these confounders on each of the outcomes of interest was clearly highlighted in the different logistic regression models outlined in this article (which is information lacking in a similar study evaluating 30-day postoperative outcomes for IMS compared to IMC patients with CD).22 While nutritional status, an important variable that has been associated with postoperative complications in CD, was not available, we used preoperative weight loss and albumin level to gauge the overall nutritional status of our patient population.30 Furthermore, we used stringent inclusion and exclusion criteria to create a well-characterized study population. By excluding those who required emergency surgery, our study population was made more homogeneous ultimately representing patients with CD who underwent elective surgery.

Nonetheless, several limitations from this study merit attention. Though this is a large multicenter database, it remains a limited sample which makes the subgroup analyses on the outcome of leak for patients without a diverting stoma underpowered thus limiting the conclusions drawn in this subset of patients. In addition, we were unable to evaluate the role of different kinds of immunosuppressant medications individually as this information was not available in this database. Patients with CD are often on multimodal agents and the risk of postoperative wound infections cannot be attributed to a specific immunosuppressant medication. Furthermore, whether the patients had adequate drug levels at the time of operation to actually cause immunosuppression is also unknown. Lau et al. evaluated 30-day postoperative outcomes in patients with CD and ulcerative colitis (UC) who underwent abdominal surgery by a single surgeon and analyzed their level of anti-TNF-α levels from stored serum samples within 7 days prior to their operation.7 Almost 50% of patients treated with anti-TNF-α therapy preoperatively failed to have detectable anti-TNF-α levels at the time of surgery. In patients with CD, an increased rate of overall postoperative infectious complications and readmissions was found in the detectable serum anti-TNF-α drug level group. Although they showed a dose relationship with higher levels of detected anti-TNF-α drug levels being associated with higher rates of adverse postoperative outcomes, their results were not statistically significant.7 The paucity of data regarding the severity and extent of CD in our patient population is another limitation. Patients with complex CD including fistulas and abscesses are at greater risk of OSI; however, this could not be accounted for in the database.30 Thus, while we attempted to adjust for potential confounders when reporting postoperative complications, we acknowledge that we might not have accounted for all of them. This, too, applies to intraoperative and postoperative antibiotic use, as these details were also not available. In a similar vein, we were unable to clearly account for the presence of a fistula. Thus, “wound classification” which we adjusted for in the multivariate analysis was used as a surrogate due to its high correlation with this variable. Along those lines, another limitation to be highlighted is the possible selection bias between patients who are on immunosuppressant medications and those who are not. While it remains probable that some patients may have been on immunosuppressive agents but had their surgery delayed because of the timing of administration of these agents, the inability to accurately account for patient’s disease severity and the indication for immunosuppression should be noted.

Conclusion

There is currently no consensus on the perioperative management of patients with CD on immunosuppressive agents. In this multicenter, nationally validated database with rigorous, standardized data collection, a significantly increased rate of infectious complications, anastomotic leak, SSI, and OSI was observed following colectomies in IMS patients with CD compared to IMC patients. When possible, consideration for a laparoscopic approach and allowing a washout period for these agents prior to surgery may decrease this associated risk. Given the inability to accurately study the effect of diversion on decreasing the risk of postoperative complications because of sample size limitations, we are unable to specifically recommend diversion or not in this patient population. However, as we observed higher rates of leak and OSI in the IMS, consideration for diversion in this group is important to mitigate the clinical sequelae of these complications.31,32 Prospective studies are needed to better design recommendations on the need for preoperative discontinuation of immunosuppressive agents in patients with CD. Furthermore, future studies are needed to determine the risk of complications due to each type of immunosuppressive agent. IMS patients should be referred to tertiary care centers with specialized IBD teams for a multidisciplinary approach and access to minimally invasive techniques.